Spring 2014 18.440 Final Exam Solutions
1. (10 points) Let X be a uniformly distributed random variable on [−1, 1].
(a) Compute the variance of X 2 . ANSWER:
Var(X 2 ) = E[(X 2 )2 ] − E[X 2 ]2 ,
and
E[X 2 ] =
Z
1
−1
2 2
4
Z
1
E[(X ) ] = E[X ] =
−1
so
Var(X 2 )
=
x3 1
= 1/3,
6 −1
x5 1
x4
dx =
= 1/5,
2
10 −1
(x2 /2)dx =
E[(X 2 )2 ] − E[X 2 ]2
= 1/5 − (1/3)2 = 1/5 − 1/9 = 4/45.
(b) If X1 , . . . , Xn are independent copies of X, and
Z = max{X1 , X2 , . . . , Xn }, then what is the cumulative distribution
function FZ ? ANSWER: FX1 (a) = (a + 1)/2 for a ∈ [−1, 1]. Thus

a+1 n

a ∈ [−1, 1]
 2
FZ (a) = FX1 (a)FX2 (a) . . . FXn (a) = 0
a < −1


1
a>1
2. (10 points) A certain bench at a popular park can hold up to two
people. People in this park walk in pairs or alone, but nobody ever sits
down next to a stranger. They are just not friendly in that particular way.
Individuals or pairs who sit on a bench stay for at least 1 minute, and tend
to stay for 4 minutes on average. Transition probabilities are as follows:
(i) If the bench is empty, then by the next minute it has a 1/2 chance of
being empty, a 1/4 chance of being occupied by 1 person, and a 1/4
chance of being occupied by 2 people.
(ii) If it has 1 person, then by the next minute it has 1/4 chance of being
empty and a 3/4 chance of remaining occupied by 1 person.
(iii) If it has 2 people then by the next minute it has 1/4 chance of being
empty and a 3/4 chance of remaining occupied by 2 people.
(a) Use E, S, D to denote respectively the states empty, singly occupied,
and doubly occupied. Write the three-by-three Markov transition
matrix for this problem, labeling columns and rows by E, S, and D.
ANSWER:


1/2 1/4 1/4
1/4 3/4 0 
1/4 0 3/4
(b) If the bench is empty, what is the probability it will be empty two
minutes later? ANSWER: 12 21 + 14 14 + 14 14 = 6/16 = 3/8.
(c) Over the long term, what fraction of the time does the bench spend
in each of the three states? ANSWER: We know


1/2
1/4
1/4
E S D 1/4 3/4 0  = E S D
1/4 0 3/4
and E + S + D = 1. Solving gives E = S = D = 1/3.
3. (10 points) Eight people throw their hats into a box and then randomly
redistribute the hats among themselves (each person getting one hat, all 8!
permutations equally likely). Let N be the number of people who get their
own hats back. Compute the following:
(a) E[N ] ANSWER: 8 ×
1
8
=1
(b) P (N = 7) ANSWER: 0 since if seven get their own hat, then the
eighth must also.
(c) P (N = 0) ANSWER: This is an inclusion exclusion problem. Let
Ai be the event that the ith person gets own hat. Then
P (N > 0) = P (A1 ∪ A2 ∪ A3 ∪ . . . ∪ A8 )
X
X
X
=
P (Ai ) −
P (Ai ∩ Aj ) +
P (Ai ∩ Aj ∩ Ak ) − . . .
i
i<j
i<j<k
8 1
8 1
8
1
=
−
+
...
1 8
2 8·7
3 8·7·6
= 1/1! − 1/2! + 1/3! + . . . − 1/8!
Thus,
P (N = 0) = 1−P (N > 0) = 1−1/1!+1/2!−1/3!+1/4!+1/5!−1/6!+1/7!−1/8! ≈ 1/e.
4. (10 points) Suppose that X1 , X2 , X3 , . . . is an infinite sequence of
independent random variables which are each P
equal to 5 with probability
1/2 and −5 with probability 1/2. Write Yn = ni=1 Xi . Answer the
following:
(a) What is the probability that Yn reaches 65 before the first time that
it reaches −15? ANSWER: Yn is a martingale, so by the optional
stoppingtheorem, we have E[Y
T ] = Y0 = 1 (where
T = min n : Yn ∈ {−15, 65} ). We thus find
0 = Y0 = E[YT ] = 65p + (−15)(1 − p) so 80p = 15 and p = 3/16.
(b) In which of the cases below is the sequence Zn a martingale? (Just
circle the corresponding letters.)
(i) Zn = 5Xn
Q
(ii) Zn = 5−n ni=1 Xi
Q
(iii) Zn = ni=1 Xi2
(iv) Zn = 17
(v) Zn = Xn − 4
ANSWER: (iv) only.
5. (10 points) Suppose that X and Y are independent exponential random
variables with parameter λ = 2. Write Z = min{X, Y }
(a) Compute the probability density function for Z. ANSWER: Z is
exponential with parameter λ + λ = 4 so FZ (t) = 4e−4t for t ≥ 0.
(b) Express E[cos(X 2 Y 3 )] as a double integral. (You don’t have to
explicitly
the integral.) ANSWER:
R ∞ R ∞ evaluate
2 y 3 ) · 2e−2x · 2e−2y dydx
cos(x
0
0
6. (10 points) Let X1 , X2 , X3 be independent standard die rolls (i.e., each
of {1, 2, 3, 4, 5, 6} is equally likely). Write Z = X1 + X2 + X3 .
(a) Compute the conditional probability P [X1 = 6|Z = 16]. ANSWER:
One can enumerate the six possibilities that add up to 16. These are
(4, 6, 6), (6, 4, 6), (6, 6, 4) and (6, 5, 5), (5, 6, 5), (5, 5, 6). Of these, three
have X1 = 6, so P [X1 = 6|Z = 16] = 1/2.
(b) Compute the conditional expectation E[X2 |Z] as a function of Z (for
Z ∈ {3, 4, 5, . . . , 18}). ANSWER: Note that
E[X1 + X2 + X3 |Z] = E[Z|Z] = Z. So by symmetry and additivity
of conditional expectation we find E[X2 |Z] = Z/3.
7. (10 points) Suppose that Xi are i.i.d. uniform random variables on [0, 1].
(a) Compute the
generating function for X1 . ANSWER:
R 1 moment
et −1
tX
tx
1
E(e ) = 0 e dx = t .
P
(b) Compute the moment generating function for the sum ni=1 Xi .
n
t
ANSWER: e −1
t
8. (10 points) Let X be a normal random variable with mean 0 and
variance 5.
R∞
2
1
(a) Compute E[eX ]. ANSWER: E[etX ] = −∞ √5√
e−x /(2·5) e−x dx.
2π
A complete the square trick allows one to evaluate this and obtain
e5/2 .
(b) Compute E[X 9 + X 3 − 50X + 7]. ANSWER:
E[X 9 ] = E[X 7 ] = E[X] = 0 by symmetry, so
E[X 9 + X 3 − 50X + 7] = 7.
9. (10 points) Let X and Y be independent random variables. Suppose X
takes values {1, 2} each with probability 1/2 and Y takes values {1, 2, 3}
each with probability 1/3. Write Z = X + Y .
(a) Compute the entropies H(X) and H(Y ). ANSWER:
H(X) = −(1/2) log 12 − (1/2) log 12 = − log 12 = log 2. Similarly,
H(Y ) = −(1/3) log 31 − (1/3) log 13 − (1/3) log 31 = − log 13 = log 3.
(b) Compute H(X, Z). ANSWER:
H(X, Z) = H(X, Y ) = H(X) + H(Y ) = log 6.
(c) Compute H(2X 3Y ). ANSWER: Also log 6, since each distinct
(X, Y ) pair gives a distinct number for 2X 3Y .
10. (10 points) Suppose that X1 , X2 , X3 , . . . is an infinite sequence of
independent random variables which are each equal to Q
2 with probability
1/3 and .5 with probability 2/3. Let Y0 = 1 and Yn = ni=1 Xi for n ≥ 1.
(a) What is the the probability that Yn reaches 4 before the first time
1
that it reaches 64
? ANSWER: Yn is a martingale, so by the
optional stopping theorem, E[YT ] = Y0 = 1 (where
T = min n : Yn ∈ {1/64, 4} ). Thus E[YT ] = 4p + (1/64)(1 − p) = 1.
Solving yields p = 63/255 = 21/85.
(b) Find the mean and variance of log Y400 . ANSWER: log X1 is log 2
with probability 1/3 and − log 2 with probability 2/3. So
E[log X1 ] =
1
2
− log 2
log 2 + (− log 2) =
.
3
3
3
Similarly,
2
1
E[(log X1 )2 ] = (log 2)2 + (− log 2)2 = (log 2)2 .
3
3
Thus,
Var(X1 ) = E[(log Xi )2 ]−E[log Xi ]2 = (log 2)2 −
− log 2 2
1
8
= (log 2)2 (1− ) = (log 2)2 .
3
9
9
Multiplying, we find E[log Y400 ] = 400E[log X1 ] = −400(log 2)/3.
And Var[log Y400 ] = (3200/9)(log 2)2 .
(c) Compute EY100 . ANSWER: Since Yn is a martingale, we have
E[Y100 ] = 1. This can also be derived by noting that for independent
random variables, the expectation of a product is the product of the
expectations.